From af2228cf3f3c50dab7f4705e09f546a53f2ee999 Mon Sep 17 00:00:00 2001
From: Jordan <me@jordan.im>
Date: Wed, 2 Dec 2020 20:56:20 -0700
Subject: initial commit

---
 README   |  7 +++++++
 pouch.sh | 47 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)
 create mode 100644 README
 create mode 100755 pouch.sh

diff --git a/README b/README
new file mode 100644
index 0000000..39dac0f
--- /dev/null
+++ b/README
@@ -0,0 +1,7 @@
+Usage: pouch.sh HTML_FILE_PATH
+
+      HTML_FILE is a pocket-exported document containing the set of URLs to save
+
+Dependencies:
+
+      google-chrome 59+ (headless mode support)
diff --git a/pouch.sh b/pouch.sh
new file mode 100755
index 0000000..500a3a5
--- /dev/null
+++ b/pouch.sh
@@ -0,0 +1,47 @@
+#!/usr/bin/env bash
+
+PROGRAM="${0##*/}"
+
+cmd_usage() {
+    cat >&2 <<-_EOF
+    Usage: $PROGRAM HTML_FILE_PATH
+
+      HTML_FILE is a pocket-exported document containing the set of URLs to save
+
+    Dependencies:
+
+      google-chrome 59+ (headless mode support)
+    _EOF
+}
+
+get_urls() {
+    sed -n 's/.*href="\([^"]*\).*/\1/p' $1
+}
+
+url_to_filename() {
+    echo "${1##*//}" |        # remove protocol (https://...)
+        sed 's/\/$//' |       # remove trailing slash
+        tr /. - |             # replace /. characters with -
+        tr -cd '[[:alnum:]-]' # remove non-alphanumeric/"-" chars
+}
+
+save() {
+    read url
+    pdfname=$(url_to_filename $url).pdf
+
+    if [[ ! -f ./$pdfname ]]; then
+        echo [+] $url...
+        google-chrome --headless --disable-gpu --print-to-pdf=$pdfname $url
+    fi
+}
+
+if [[ $# -eq 1 && ( $1 == --help || $1 == -h || $1 == help ) ]]; then
+    cmd_usage
+elif [[ $# -eq 1 ]]; then
+    get_urls $1 | while read line ; do save $line ; done
+else
+    cmd_usage
+    exit 1
+fi
+
+exit 0
-- 
cgit v1.2.3-54-g00ecf